For reducing data so that the terrestrial albedo can be determined we have forward and inverse methods.
1) The BBSO linear sky-fit method. Yields a cleaned-up DS so that DS/BS ratio can be found and used for the inversion. Probably works best near the edge of the DS as the scattered light is not a linear function of distance except far from the BS.
2) Andrew’s clean-up method based on fitting a model image convolved with a variable PSF until it can be subtracted perfectly from the sky-part of an observed image. As a model of the Moon an observation – minus the sky part and the DS – is used.
3) A synthetic model of the Moon is used within a convolution framework to produce an image that ‘looks just like the Moon and its halo’. Since a parameter in the method is the terrestrial SSA it is directly determined in this way, but does depend on the same assumptions about time, geometry, lunar reflectance and so on as does the inverting methods. Depending on how much of the lunar disc is included in the fit there may be a difference in how the knowledge of lunar albedo and lunar reflectance influences bias.